How to learn imbalanced data arising from multiple domains

On Multi-Domain Long-Tailed Recognition, Imbalanced Domain Generalization and Beyond (ECCV 2022)

Let me introduce you to our new work, which has been accepted by ECCV 2022: On Multi-Domain Long-Tailed Recognition, Imbalanced Domain Generalization and Beyond. As the name suggests, the problem of this work is how to learn a robust model when there are data from multiple domains, and each of the domains can have (potentially different) data imbalance. Existing methods for dealing with imbalanced data/long-tailed distribution are only …

artificial intelligence computer vision data deep learning editors pick imbalanced-data